How we build our replicas

How a replica is built

Every replica assessment on the platform is built to the provider's published specification. The starting point is always material the provider has published itself — fact sheets, candidate guides and official practice pages from SHL, Aon and Cappfinity. From that material, each replica is built to match:

• Timings — the overall time limit and the pace per question it implies.
• Question formats — the item types, answer mechanics and on-screen layout conventions of the real test.
• Scoring rules — including negative marking where the provider uses it.
• Navigation behaviour — whether the real test allows you to go back to earlier questions.

The source standard

Provider-published material outranks everything else. Prep-industry sites make many claims about these tests, and some are wrong — so a prep-industry claim is never, on its own, the basis for anything in a replica. Where a provider publishes nothing on a detail, we make a conservative choice rather than presenting a guess as fact.

How the Benchmark works

The Benchmark is a timed, 20-question reasoning assessment built in the style of the online tests used by UK finance and consulting employers. It covers three sections — numerical, verbal and logical reasoning — in 22 minutes, under realistic time pressure.

It is free, and the results are free: your score, your estimated percentile, your per-section strengths and weaknesses, and an answer review with the correct answer for every question. There is no paywall between you and your results.

You only get one attempt — deliberately. Most practice tools give you unlimited attempts. That's useful for practising, but it means your score stops being information: retake anything often enough and the number tells you about your familiarity with that test, not your ability. The Benchmark is the opposite instrument — one attempt, taken cold. That single constraint is what makes the result an honest baseline. Practice belongs after the diagnosis, not inside it.

Your answers are scored server-side against a fixed answer key — the score you see is computed and stored by our system, not your browser, so it can't be inflated or edited. You get an overall score and a per-section breakdown, so you can see not just where you stand but where your preparation time will pay off most.

How the estimated percentile is produced

The percentile next to your score is an estimate from a statistical model calibrated to graduate-level reasoning performance.

We label it "estimated" everywhere it appears, deliberately: a model is not a measurement. As real candidate cohorts accumulate, the estimate will be increasingly informed by real results — but we will never present a modelled figure as if it were a direct count of people you beat. If a prep tool shows you a percentile without telling you where it comes from, that's worth asking about.

Corrections policy

Providers change their tests — formats are retired, timings revised, new versions rolled out. When a provider's published specification changes, the affected replica is corrected to the new specification. And if we find that a replica deviates from what the provider has published, correcting it takes priority over building new content.

What we do with results

Individual results are yours. In aggregate, de-identified results are used to calibrate and refine the percentile model. Full details are in our privacy policy.